To get your law firm visible to AI assistants, make the site machine-readable, rewrite key pages so answers sit in the first forty words, keep your firm details identical everywhere, then earn citations from sources those assistants already trust. The rest of this guide shows exactly how - with a quick-wins checklist, copy-paste audit prompts, and how ChatGPT, Google AI Overviews, Perplexity, Copilot, and Gemini differ.
Quick wins you can finish this week
Most firms that are invisible to AI assistants fail on basics, not on exotic tactics. Work these in order before you rewrite your entire site.
- Confirm AI crawlers are allowed. Check
robots.txtfor GPTBot, ChatGPT-User, OAI-SearchBot, PerplexityBot, and ClaudeBot. Accidental blocks are common after security plugins or default configs. - Ship structured data on the firm and each attorney. LegalService for the firm, Attorney for each lawyer, FAQPage on any real Q&A block. Models prefer being told what a page is over guessing.
- Rewrite the first forty words on your top three money pages. Lead with a direct answer a client would recognize, then expand. Brochure openers do not get quoted.
- Align name, address, phone, and practice areas across your site, Google Business Profile, and major directories. One spelling, one format, everywhere.
- Write down ten client prompts and run them monthly in ChatGPT, Perplexity, and Copilot. Record whether you are named and which sources were cited instead - that list is your outreach map.
What "visible to AI assistants" actually means
Visibility here is not a ranking position. It is whether an assistant names your firm inside an answer when someone asks for a lawyer, a next step after an accident, or help with a filing. There is usually no list of ten blue links and often no click. You are either in the answer or you are not.
That is why traditional SEO still matters and is no longer enough on its own. Pages that already rank are more likely to be retrieved. Pages that answer plainly, carry consistent entity data, and are corroborated elsewhere are more likely to be quoted. The work overlaps; the measurement does not. Rankings are measured by position. AI visibility is measured by citation.
LexGrow's own tracking makes how early this market still is concrete. Between 11 June and 9 September 2026, LexGrow earned 55 Microsoft Copilot citations, and on one commercial query held a 21 percent share of the sources Copilot grounded its answer on. Those are small absolute numbers - which is the point. The firms competing for them are few, and the cost of entry is structure and authority, not ad spend.
The four-step sequence that actually works
Sequence matters more than volume. A firm that earns credible references and writes plainly will be cited ahead of a firm that published two hundred pages nobody links to.
Step 1: Make the site readable by a machine
Before anything else, confirm an assistant can actually read the page.
- Fetch each key page as a crawler would and confirm the content is in the raw HTML, not injected only after JavaScript runs.
- Publish LegalService, Attorney, and FAQPage markup where it applies.
- Allow the AI crawlers listed above in
robots.txt. - Publish an
llms.txtat the site root that describes what the firm does and links the pages worth reading.
This is the cheapest work on the list and the most commonly skipped. It gates everything else.
Step 2: Rewrite the pages that answer real questions
Assistants quote passages, not pages. A thorough page can still never be quoted if no paragraph stands alone once lifted out of context.
- Put the answer in the first forty words. No preamble, no credentials dump, no story first.
- Write headings as the question a client would type, in their words rather than legal terms.
- Keep paragraphs self-contained. Assume any one of them might be quoted alone.
- Use checkable specifics: jurisdictions, filing deadlines, statute names, dollar ranges. Adjectives do not survive verification.
Step 3: Make your details agree everywhere
An assistant that cannot confirm a detail across sources will usually drop it rather than risk it. The most common failure is a firm whose own site disagrees with its Google Business Profile, or whose name is spelled three ways across directories.
Pick one canonical firm name, one address format, and one phone number, then make every profile match exactly. Bar directory listings and the Business Profile matter more here than any page you control, because they are often what a model checks you against.
Step 4: Earn citations - the slow part that decides it
Steps one through three remove reasons not to cite you. Step four is what gets you cited, and it is measured in quarters rather than weeks. There is no version of this that skips it.
- Get referenced by sites that already get quoted in your practice area and city: local press, bar associations, legal publications, university clinics.
- Publish something worth referencing - original data, a clear explainer of a local rule change, a genuinely useful template - rather than something written only to rank.
- Keep reviews current across platforms. Recency counts; a stale profile underperforms a thinner but active one.
How the major assistants differ
Treating "AI search" as one channel is the most common mistake. The systems differ in where they look and what they will quote. A firm can be strong in one and absent from another.
| Assistant | Where it looks | What it favours | What that means for a firm |
|---|---|---|---|
| ChatGPT Search | Live web index plus training data | Pages that answer directly, with sources it can attribute | Answer-first structure matters more than keyword coverage |
| Google AI Overviews | Google's own index | Pages already ranking, with clear structure and schema | Conventional ranking is the entry ticket; structure decides whether you are quoted |
| Perplexity | Live web, citations shown inline | Recent, specific, well sourced pages | Publication date and named specifics carry unusual weight |
| Microsoft Copilot | Bing's index | Pages Bing has crawled and rates as authoritative | Bing Webmaster Tools coverage is worth checking separately from Google |
| Gemini | Google index plus Google properties | Corroborated entities with a consistent profile | Business Profile accuracy feeds this one more than the others |
For the full comparison and how LexGrow measures citation share across assistants, see the AI visibility hub.
Three prompts to audit your own AI visibility
Copy these into ChatGPT, Perplexity, and Copilot. Swap in your city and practice area. Run the same set monthly and keep a spreadsheet of who gets named.
- Recommendation prompt: "I need a [practice area] lawyer in [city, state]. Who would you recommend and why? List the sources you used."
- Comparison prompt: "Compare the top [practice area] firms in [city] for someone who just [common client situation]. Which firms appear most often in reliable sources?"
- Gap prompt: "What should I look for when choosing a [practice area] attorney in [city]? Which local firms are frequently cited for that advice?"
When you are not named, the useful output is the list of sources the assistant cited instead. That list is your Step 4 outreach target: the sites already trusted enough to be quoted.
What this is not
There is no submission form, no paid placement inside ChatGPT or Google AI Overviews, and no way to make an assistant name you on demand. Google may show ads adjacent to AI Overviews; that is advertising next to the answer, not inclusion within it. Anyone selling guaranteed placement inside a generative answer is selling something that does not exist.
What you can measure is whether your firm currently appears for the prompts your clients actually use, which sources the assistant quoted instead, and whether that changes as the structural and authority work lands.
How long it usually takes
Structural work - schema, crawlability, answer-first rewrites - can show within weeks of a recrawl, because it only requires the assistant to re-read pages it can already reach. Being named in answers where you currently are not usually takes quarters, because that depends on other sites referencing you. A firm with an established site and current reviews tends to move sooner than a new domain, for the same reasons it ranks sooner.
Where to go next
If you want the deeper playbook - per-assistant differences, what actually moves a citation, and how to measure it - start on the LexGrow AI visibility page. If you want the ordered checklist without the theory, use how to appear when people search in AI. For the broader framing of how generative search relates to traditional search, see AI search and law firm visibility.
LexGrow helps law firms earn citations from ChatGPT, Google AI Overviews, Perplexity, and Copilot - then measure whether it is working. Book a free visibility review or run the AI visibility audit.